Internship | Applying deep learning to time series of radar data

What will be your role?

Nowadays, deep learning techniques are used in a wide variety of domains, including the radar domain. In the radar domain, deep learning is primarily applied for classification based on some 2D representation of the radar data, e.g., an Inverse Synthetic Aperture Radar (ISAR) image or a spectrogram (i.e., an image of signal frequency content versus time). These intermediate data representations are commonly used because they are easily interpretable for humans. Depending on the final application, however, deep neural networks might be trained more efficiently using other data representations. For specific applications it might even be beneficial to omit the intermediate data representation and train a neural network using pulse-compressed radar data or directly raw radar data.

The focus of this assignment is the use of pulse-compressed or raw radar data for training a deep neural network. What is the potential of this approach? W...